Shared Queue System for PCIe Traffic Resource Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computer bus systems face inefficiencies due to increased queue sizes and latency as data rates scale, leading to higher silicon real estate requirements and power consumption, particularly in PCIe environments, where multiple virtual channels and flow-control classes result in underutilized resources and increased latency.
Innovation Solution
Implementing a shared queue system that pools credits across flow-control classes and virtual channels, allowing for resource sharing and optimized credit management to reduce overall queue size requirements, while maintaining PCIe ordering rules and resilience against DLLP corruptions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If separate queues are allocated for each virtual channel and flow-control class, then communication reliability and ordering rules are maintained, but silicon area and power consumption increase due to larger queue sizes
Solution Approach 1:
The patent merges multiple separate queues (one for each virtual channel and flow-control class combination) into a single shared queue structure. This consolidation allows the system to maintain the required PCIe ordering rules and resilience against DLLP corruptions while significantly reducing the total silicon area required, as the shared queue eliminates redundant queue memory across multiple virtual channels and flow-control classes.
Solution Approach 2:
The shared queue is designed to serve multiple virtual channels and flow-control classes simultaneously, making it a universal resource that replaces multiple dedicated queues. This multi-functional queue structure maintains the necessary communication reliability for different traffic types while reducing overall queue size requirements and silicon real estate.
2Reliability
If separate queues are allocated for each virtual channel and flow-control class, then communication reliability is maintained, but power consumption increases due to larger queue sizes
Solution Approach 1:
The patent merges multiple separate queues into a single shared queue structure, which directly reduces power consumption by eliminating the energy required to maintain and manage multiple redundant queue structures. The shared queue maintains the necessary PCIe ordering rules and resilience against DLLP corruptions while consuming less power due to its smaller overall size.
Solution Approach 2:
The patent changes the structural parameters of the queue system from multiple large queues to a single shared queue with optimized size. This parameter change reduces the total memory capacity required, which directly lowers power consumption associated with queue maintenance, while preserving the reliability requirements for PCIe ordering and DLLP corruption resilience.
3Productivity
If queue sizes are increased to handle multiple virtual channels and flow-control classes, then bandwidth capacity is maintained, but latency increases due to larger queue depths
Solution Approach 1:
The patent segments the shared queue into multiple virtual channels and flow-control classes through logical partitioning rather than physical separation. This segmentation allows the system to maintain full bandwidth capacity across all channels while keeping individual queue depths smaller, thereby reducing latency compared to a single large monolithic queue structure.
Solution Approach 2:
The patent resolves the bandwidth-latency tradeoff by introducing a new dimensional organization of the queue system. Instead of increasing queue depth in one dimension to handle multiple channels, the system uses a multi-dimensional shared queue structure that allows concurrent access across virtual channels while maintaining shallow individual queue depths, thus preserving both bandwidth and reducing latency.
4Reliability
If multiple separate queues are used for different traffic types, then traffic isolation and ordering are maintained, but resource utilization decreases leading to underutilized resources
Solution Approach 1:
The patent merges multiple dedicated queues into a single shared queue while maintaining logical isolation through virtual channel and flow-control class identifiers. This merging enables dynamic resource sharing where idle queue space from one virtual channel or flow-control class can be utilized by other channels, significantly improving resource utilization while preserving traffic ordering and isolation requirements.
Solution Approach 2:
The patent introduces dynamic resource allocation within the shared queue, allowing queue resources to be flexibly assigned and reassigned based on current traffic demands across different virtual channels and flow-control classes. This dynamic approach contrasts with static dedicated queues and enables better resource utilization while maintaining the necessary traffic isolation and ordering through logical separation mechanisms.
Data Source
AI summary
Systems, methods, and computer-readable media are disclosed for an apparatus coupled to a communication bus, where the apparatus includes a queue and a controller to manage operations of the queue. The queue includes a first space to store a first information for a first traffic type, with a first flow class, and for a first virtual channel of communication between a first communicating entity and a second communicating entity. The queue further includes a second space to store a second information for a second traffic type, with a second flow class, and for a second virtual channel of communication between a third communicating entity and a fourth communicating entity. The first traffic type is different from the second traffic type, the first flow class is different from the second flow class, or the first virtual channel is different from the second virtual channel. Other embodiments may be described and/or claimed.